When sampling full-sibs for linkage studies, half-sibs are often available. A theoretical method is proposed to combine the information from full-sib and half-sib pairs into a single test for linkage. The method is based on the Haseman and Elston method of regressing the squared trait-difference for a pair of sibs (either full- or half-sibs) on the estimated proportion of alleles shared identical by descent. A correction factor that consideres the correlation among sibs is proposed to approximate the distribution of the test statistic; simulations are used to demonstrate that the approximation works well in many situations, although there are some situations for which the statistic can have an inflated type-I error. - computer simulation, G.A.S.P., genometrics, sib-pair linkage analysis, half-sibs